PDA

View Full Version : Re: genlocking digital camcorders and software genlock summery

Young-hoo Kwon, Ph.d.
11-10-2003, 03:04 AM
Dear Kevin and all:

I posted an inquiry sometime ago asking software genlock references.
Tomislav Pribanic sent me a JB article by Pourcelot et al.:

Pourcelot et al. (2000). A method to synchronise cameras using the
direct linear transformation technique. J Biomechanics 33, 1751-1754.

This was what I was looking for. Thank you, Tomislav!

The software genlock technique based on the DLT method is using the
error involved in the point reconstruction. DLT method is no different
from the regular linear multiple regression and one can compute the MSE
involved in the least square estimation. Imagine two lines in the space
not parallel but not crossing each other either. One can draw a third
line crossing both lines and perpendicular to both which will show the
shortest distance between the two lines. The least square solution of
the reconstructed point is around the mid point of the third line. (It
is not exactly the mid point of the line because the minimization is
done based on the image plane coordinates, not the object space
coordinates.) Anyway this is the situation when the cameras are not
genlocked. The lines of sight from the cameras are not crossing each
other due to the time error. By systematically introducing time offsets
among the cameras, one can shorten the third line. Finding the time
offsets that minimizes MSE is what the software genlock is all about.

I recently applied this approach to a golf swing study and compared the
optimized time offsets with the measured time offsets (4 Panasonic
Proline digital camcorders). The time error was about 1/100 to 2/100 of
the field interval (1/60 s) or 0.1 - 0.3 ms. The MSE was reduced
substantially in this process.

Now back to Kevin's question, I received another email from Thomas
Seeholzer of SIMI. It sounds like the SIMI system solves this problem at
the time of recording. SIMI computes the timeoffsets at the time of
recording. I am not quite sure whether they used the software genlock
method described above or something else.

It appears to me that the software genlock method is the only solution
for the digital camcorders at the moment. In terms of programming, all
you need to do is to identify the markers to be used in the software
genlock and determine the frame range. The bottom line is you have to
use fast-moving points and phase for this purpose because fast-moving
points will easily magnify the MSE due to time error. Compute the mean
MSE of the markers and frames. As you introduce time offsets
systematically, identify the minimum mean MSE conditions. That is the
optimized time offset.

I hope it helped.

Young-Hoo Kwon
------------------------------------------------------
- Young-Hoo Kwon, Ph.D.
- Biomechanics Lab, Texas Woman's University
- kwon3d@kwon3d.com
- http://kwon3d.com
------------------------------------------------------

-----Original Message-----
From: Biomechanics and Movement Science listserver
[mailto:BIOMCH-L@NIC.SURFNET.NL] On Behalf Of Kevin Arnold
Sent: Monday, November 10, 2003 8:24 AM
To: BIOMCH-L@NIC.SURFNET.NL
Subject: [BIOMCH-L] genlocking digital camcorders

Does anyone know of a way in which to genlock digital camcorders? We're
urgently needing to synchronise our digital camcorders (Sony
TRV900E) but haven't been able to find any cable capable of doing so.
Does anyone have any suggestions towards what makes and models of
camcorders to use, and whereabouts to find the cables for connecting the
cameras.

Kevin

------------------------------------------------------------------------
------------------
Kevin Arnold
Biomechanics Technician
Department of Sport & Exercise Science
University of Portsmouth
St Michaels Building
Portsmouth, PO1 2DT
Tel: 023 9284 2028
kevin.arnold@port.ac.uk

-----------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
Web-based discussion forum: http://movement-analysis.com/biomch_l
-----------------------------------------------------------------

-----------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
Web-based discussion forum: http://movement-analysis.com/biomch_l
-----------------------------------------------------------------